Section: New Results

Control techniques

Participants : Francisco Navas Matos, Carlos Eduardo Flores Pino, David Gonzalez Bautista, Joshué Pérez Rastelli, Vicente Milanés Montero.

The final stage for automating a vehicle relies on the control algorithms. They are in charge of providing the proper behavior and performance to the vehicle, leading to provide the fully automated capabilities. Having this in mind, there are two research lines currently open in the time: the first one is mainly related to what we call “naturalistic driving” in the sense of adding the human reasoning to the vehicle. We are mainly focusing our effort on artificial intelligent algorithms as neuro-fuzzy techniques. The main reason is the growing interest of the car makers in adding sharing control capabilities (between the vehicle and the driver) to the automated car. Our initial results show a big potential of using this approach and we already achieved some simulations results that were well-accepted by the scientific community and will be shown in mid-December at the final event of the EU project DESERVE.

On the other hand, we are also further investigating robust control algorithms for providing stability not only to an automated vehicle but also to a chain of automated vehicles that should be able to cooperate intelligently.

This work is mainly divided in two main research lines:

1) Controllability and stability of dynamic complex systems are the key aspects when it comes to design intelligent control algorithms for vehicles. Current advances in the field are mainly oriented to advanced multi-sensor fusion toward multi-target decision-making systems. These artificial intelligence-based algorithms are able to provide reasonable responses under controlled environments (i.e. highly-detailed maps). However, new trends are proposing intelligent algorithms able to handle any unexpected circumstances as unpredicted uncertainties or even fully outages from sensors. The goal of this new research line at RITS is to further investigate control algorithms able to provide stability responses for autonomous vehicles under uncontrolled circumstances, including modifications on the input/output sensors. Dynamic plant models where different inputs/outputs can be added or subtracted in real-time during its operation is one of the hot topics in the control research arena. This system has to provide stable enough response when these operations occur. This is especially true on high-risk environments as autonomous driving; and

2) Data-driven control techniques based on model-free algorithms. Vehicles exhibit a highly non-linear behavior, especially at low speeds (as occur in urban environments). The research on novel data-driven techniques that are independent of the plant model provides huge benefits when applying them to automated vehicles. This novel research line in the team tries to further investigate on stable algorithm that doesn’t need an accurate model of the vehicle dynamic, leading to compensate the effects of nonlinear dynamics, disturbances, or uncertainties in the parameters. [35]